On 18.06.2013, at 18:04, Fabien Chouteau wrote:
> On 06/18/2013 05:31 PM, Alexander Graf wrote:
>>> + for (type = 0; type < 2; type++)
>>
>> You need braces on these. Please run your patch through checkpatch.pl :).
>
> I did ;)
>
> -> ./scripts/checkpatch.pl 0001-PPC-Add-dump_mmu-for-6xx.patch
> total: 0 errors, 0 warnings, 51 lines checked
>
> 0001-PPC-Add-dump_mmu-for-6xx.patch has no obvious style problems and is
> ready for submission.
Meh - broken script :). According to the CODING_STYLE convention all of the
above need to be cluttered with braces ;).
>
>>
>>> + for (way = 0; way < env->nb_ways; way++)
>>> + for (entry = env->nb_tlb * type + env->tlb_per_way * way;
>>> + entry < (env->nb_tlb * type + env->tlb_per_way * (way +
>>> 1));
>>> + entry++) {
>>> +
>>> + tlb = &env->tlb.tlb6[entry];
>>> + cpu_fprintf(f, "TLB %02d/%02d %s way:%d %s ["
>>> + TARGET_FMT_lx " " TARGET_FMT_lx "]\n",
>>> + entry % env->nb_tlb, env->nb_tlb,
>>> + type ? "code" : "data", way,
>>> + pte_is_valid(tlb->pte0) ? "valid" : "inval",
>>> + tlb->EPN, tlb->EPN + TARGET_PAGE_SIZE);
>>> + }
>>
>> I thought 6xx and 74xx also support HTAB and SRs? Shouldn't we dump those as
>> well?
>>
>
> I don't know what that is, can you send me an example of what the printf line
> should be?
SRs are similar to the SLB that book3s_64 print out. Just that there are a
fixed smaller number of them (16). Basically you'd dump the env->sr array,
similar to how the debug functions in get_segment_6xx_tlb() dump it.
For the HTAB I think SDR1 should be enough, so you don't need to do too much
here. If you like, you can just dump the decoded fields env->htab_base and
env->htab_mask. Dumping the whole HTAB would just explode the output.
However, you also should definitely dump all (valid) BATs. Check out
get_bat_6xx_tlb() for debug code that dumps BATs.
Alex
